Estimation of Electron Transfer Distances from AM1 Calculations - The

Luisa G. Reuter , Annabell G. Bonn , A. Claudia Stückl , Bice He , Palas Baran Pati , Sanjio S. Zade , and ..... Jian Song , Fang Gao , Bo Shi , WanZ...
0 downloads 0 Views 118KB Size
J. Phys. Chem. A 2000, 104, 10023-10031

10023

Estimation of Electron Transfer Distances from AM1 Calculations Stephen F. Nelsen* Department of Chemistry, UniVersity of Wisconsin, 1101 UniVersity AVenue, Madison, Wisconsin 53706-1396

Marshall D. Newton* Department of Chemistry, BrookhaVen National Laboratory, Box 5000, Upton, New York 11973 ReceiVed: June 20, 2000

We examine a simple approximate method for calculating the electron transfer (ET) distance suitable for extracting the off-diagonal electronic coupling element (Hab) of Marcus-Hush theory from the optical spectrum of nitrogen-centered organic intervalence radical cations. A very simple estimate of the ET distance on the adiabatic ground-state surface (d12(dm)) is employed. AM1-UHF calculation of the dipole moment component in the long axis direction (µ1) for the radical cation using the center of mass as the origin gives the estimated d12(dm) (Å) ) 2µ1(Debye)/4.8023 (eq 7). Cave and Newton’s Generalized Mulliken-Hush theory equation allows calculation of the diabatic counterpart (dab) in terms of d12 and the transition dipole moment (µ12), obtained from the experimental intervalence optical band. These calculations indicate that dab is significantly smaller than the distance between the nominal sites or charge localization, ratio 0.82-0.85 for the aromaticbridged bis(hydrazines), 0.76-0.87 for the unsaturated-bridged bis(triarylamines), 0.74-0.79 for the saturatedbridged bis(diazenes), and 0.71-0.85 for the saturated-bridged bis(hydrazines) examined here. Furthermore, dab is not very different for diastereomers that differ in relative orientation of the oxidized and reduced chargebearing units for the aromatic-bridged compounds; experimental data corresponds to a superposition of the spectra of such isomers. There appears to be a problem with the trend of calculated ET distances as methyl groups are substituted on a benzene-1,4-diyl bridge. The d12(dm) calculated is smaller for the compound with the tetramethyl-substituted bridge (DU) than that with the dimethyl-substituted bridge (XY), in contrast to the general trend in d12 as twisting increases, and to the dipolar splitting constant for the triplet form of the dication oxidation states of these compounds.

Introduction Symmetrical, charge localized, intervalence (IV) compounds have two identical charge bearing units (M, often metal coordination complexes), symmetrically connected to a bridging unit (B).1 They are at an oxidation level that places different charges on the M units, so they may be symbolized as n+1M-B-Mn. They are the most revealing electron transfer (ET) systems yet devised, because they exhibit a charge-transfer (CT) band from which the ET parameters that allow prediction of the thermal electron-transfer rate constant can be estimated using MarcusHush theory.2-4 When the electronic coupling through the bridge (the off-diagonal matrix coupling element H) is small enough relative to the total vertical reorganization energy upon electron transfer (Marcus’s λ), the extra charge is mostly localized on one M unit. We will call the diabatic states (those in the hypothetical absence of electronic coupling between the Ms, that is, at H ) 0) a and b, and the adiabatic states resulting after inclusion of the electronic coupling 1 and 2. For the Mulliken-Hush model using parabolic diabatic states, the transition energy at the CT band maximum (the energy separation ν˜ max between the adiabatic surfaces at the groundstate energy minimum, often called ∆E12), is equal to the vertical energy separation at the minimum for the diabatic surfaces, ∆Eab, and is also equal to λ.2a,4d We will use Eop to denote this experimental estimate of λ.2,3 Hush related Hab to Eop using the ratio of the transition dipole moment coupling the adiabatic surfaces, µ12, to the change in dipole moment between the

ground state and excited state on the diabatic surfaces, ∆µab (eq 1).3

|Hab| ) (|µ12|/|∆µab|)Eop

(1)

Equation 1 is usually written5 using |∆µab| ) e(dab), leading to eq 2, where dab is the effective electron-transfer distance (Å),

|Hab| ) 2.06 × 10-2(max ∆ν˜ 1/2 Eop)1/2/dab

(2)

∆ν˜ 1/2 the bandwidth at half-height (cm-1), and max the extinction coefficient at the band maximum (cm-1 M-1). More sophisticated versions of eqs 1-2 derived using vibronic coupling theory (VCT)6 are available.7 They require accurate separation of λ into its classically treated solvent (λs) and quantum mechanically treated internal vibrational (λv) components (their sum, λ, is slightly larger than Eop) and also a weighted average energy of the vibrations responsible for ET (ν˜ v). Using VCT the observed IV-CT band is a superposition of transitions to each vibrational level of the excited state, weighted by the Franck-Condon factors, so the expression for bandwidth is complex. Nevertheless, a VCT treatment applied to dinitrogencentered IV compounds produces Hab values that are quite close to those using eq 2 when the same ET distances are employed,8,9 except for a refractive index (n) correction to max that was not used in obtaining eq 2. The correction corresponds to a factor in Hab of gn-1/2, i.e., a factor of g0.86 in acetonitrile at 25 °C, depending on the form of the refractive index correction used.7

10.1021/jp002211w CCC: $19.00 © 2000 American Chemical Society Published on Web 10/19/2000

10024 J. Phys. Chem. A, Vol. 104, No. 44, 2000

Nelsen and Newton

Electron-Transfer Distances. This work is concerned with how to estimate the size of dab that is appropriate to use in eq 2 applied to nitrogen-centered IV compounds. The metal-metal distance has been commonly used as dab for IV compounds with metal-centered M, because the distance in Hush’s eq 2 refers to that for the diabatic surfaces (at Hab ) 0), and if the ligands were all the same, the metal obviously would be the center of charge. However, some charge is delocalized onto the ligands, and the bridge is a ligand that is different from the others, usually more easily oxidized or reduced, so the effective dab might be argued to be somewhat smaller than the metal-metal distance. Experimental estimates of the effective d from independent measurements of ∆µ pertain to the real system, which has a nonzero Hab, and will thus be measures of d12 ≡ ∆µ12/e, although this distinction has often not been made in the literature (cf. refs 4 and 10). Charge separation distances for the excited states obtained by photoexcitation of neutral organic donorbridge-acceptor systems in nonpolar solvents have been estimated using time-resolved microwave conductivity experiments, as for Paddon-Row’s dimethoxynaphthalene, dicyanoolefin compounds linked by saturated polycyclic alkyl groups.11 Experimental evaluation of the dipole moment change upon excitation of IV compounds, and hence of their d12 values, have employed electroabsorption (Stark effect) spectroscopy.12-14 The effect of an applied electric field on the position of the absorption maximum of an IV compound frozen in a matrix is analyzed. These studies have shown that d12(Stark) may be considerably smaller than metal-metal distances, even approaching zero for strongly coupled systems.12,13 Organic charge-bearing units obviously have delocalized charges, but for convenience, the distances between the centers or edges of these units (the atoms attached to the bridge) have often been used as the dab values required to apply eq 2 for evaluation of Hab.11 In their generalized Mulliken-Hush theory,4 Newton and Cave point out that eq 3 gives the relationship between the

|∆µab| ) [(∆µ12)2 + 4(µ12)2]1/2

(3)

diabatic and adiabatic ∆µ values. Analogous to the comments made in connection with eq 1,5 eq 3 in general refers to dipole components along the charge-transfer direction. In the present study (except as noted otherwise) the latter direction corresponds to that of ∆µ b12. The ratio of d12 to dab is the ratio between the positions of the adiabatic and diabatic minima on the ET coordinate X using the simple two state model, which is 1 - 2Xmin ) (1 - 4H2/ Eop2)1/2.2a This relationship makes it easy to see that d12 approaches 0 as H approaches Eop/2 and the system changes from an instantaneously charge-localized one to one that is delocalized. The expression for µ12 obtained from eq 1 and 2 is shown in eq 4. Equation 3 may be written in terms of the ET

µ12 ) (2.06 × 10-2)e(max ∆ν˜ 1/2/Eop)1/2

(4)

Nelsen’s group has used the relatively high internal reorganization energy of nitrogen-centered charge-bearing units to increase Eop enough to allow the rate for intramolecular ET (kET) to be sufficiently slow to measure by ESR even when Hab is large enough to make the IV-CT band clearly visible.8,16 The nitrogen ESR splitting constant for hydrazine radical cations and hydrazyl radicals is in the range 11-13 G, which makes the ESR spectrum of their IV radical cations very sensitive to intramolecular electron exchange when kET is near 108 s-1. We will call kET values measured by this method kESR. The kESR value provides a sensitive check on the accuracy of the optically determined Hab value. Using Hush theory with parabolic diabatic surfaces, eq 6 with A ) 1 is the expression for the thermal ET

∆G* ) AEop/4 - Hab + Hab2/Eop

(6)

barrier, ∆G*.2 Parabolic diabatic surfaces used with the simple classical two state model of Marcus-Hush theory produce an IV-CT band having a width at half-height of ∆ν˜ 1/2HTL ) (16 RT ln(2) Eop)1/2. Many of our compounds show broader bands than this. The conventional way of handling this problem is to employ VCT using a single mean harmonic mode,6 but this procedure is insufficient for our compounds, we believe, because their high λv values require assuming harmonic energy surfaces to unreasonably high vibrational levels. We therefore assumed that the IV-CT band reveals the effective adiabatic surfaces. Using the simple two state model to relate diabatic to adiabatic surfaces requires that the diabatic surfaces are not prefect parabolas unless ∆ν˜ 1/2 ) ∆ν˜ 1/2HTL. Good fit to the observed band shape for all our compounds is obtained using a quartic term in the diabatic surfaces, that is, using Haa ) λ/(1 + C)(X2 + C X4) and similarly for Hbb, with X replaced by (1 - X) (Marcus-Hush theory uses C ) 0). This produces an increasingly smaller value of A in eq 4 as C increases. We used the ∆G* values from eq 4 in an adiabatic rate expression to obtain the rate constant for thermal ET estimated from the optical parameters of the IV-CT band, kopt.16b,c,17 There is no particular physical significance to using an X4 term in fitting the optical spectrum, and other functions fit as well.16c For compounds having saturated bridges doubly linking charge-bearing dinitrogen units (discussed later), the dinitrogen units are oriented nearly perpendicular to the electron transfer axis of the molecule. The shortest distances between the nitrogen atoms and the distance between the midpoints of the NN bonds, which ought to be very close to the centers of charge for these tri- and tetraalkyl dinitrogen systems, are nearly the same. For these compounds, using the distance between the dinitrogen units (dNN) as dab in eq 2 gives kopt values that are rather close to kESR.8,9,16 We have also studied several examples of +Hy-Ar-Hy, where Hy is the 2-tert-butyl-2,3-diazabicyclo[2.2.2]oct-3-yl unit, and Ar is a symmetrical aromatic diyl such as the ones shown below. For these compounds the NN bond axes lie at about a

distances as eq 5, which is convenient for converting d12 to dab from experimental parameters for the observed IV-CT band.

dab2 ) d122 + 4 (2.06 × 10-2)2 max ∆ν˜ 1/2/Eop

(5)

Cave and Newton showed how dab may be fully determined using either experimental (spectroscopic) or calculated data. For example, they estimated dab using eq 3 with semiempirically calculated ∆µ12 and µ12 values, employing the INDO method of Zerner and co-workers,15 at the configuration interaction (CI) level to describe the initial and final states.

120° angle to the ET axis, and the aromatic bridge is considerably different electronically from the other three (saturated alkyl) substituents at each dinitrogen unit. For +Hy-Ar-Hy, we found good agreement of kopt with kESR when we employed

ET Estimations from AM1

J. Phys. Chem. A, Vol. 104, No. 44, 2000 10025

TABLE 1: Comparison of Various Distances (Å) Calculated for Simple Aryl-Bridged IV Bis(hydrazines) compound +Hy-PH-H

+Hy-PH-Hy +

Hy-DU-Hy

+Hy-BI-Hy

dN,N a

2r(pc) b

µ12 c

dab(pc) d

d12(ESR) e

s 5.70 5.70 9.99

6.14 4.56 4.59 8.42

s 4.30 2.30 3.18

s 4.90 4.69 8.52

s 4.6-4.8 5.67 8.0

a Distance between the nitrogens bonded to the aryl bridge, from the AM1-UHF energy minimum structure (the sNsB diastereomers of these compounds). b Twice the distance from the center of charge to the center of the aryl bridge. c Transition dipole moment, in Debye. Calculated using eq 4 and the experimental optical parameters in acetonitrile: (see ref 13c) +Hy-PH-Hy Eop ) 13 000 cm-1, max ) 3800 M-1 cm-1, ∆ν˜ 1/2 ) 6460 cm-1; +Hy-DU-Hy Eop ) 14 100 cm-1, max ) 970 M-1 cm-1, ∆ν˜ 1/2 ) 7890 cm-1; +Hy-BI-Hy Eop ) 15 210 cm-1, max ) 2600 M-1 cm-1, ∆ν˜ 1/2 ) 6040 cm-1. d Calculated from µ12 in acetonitrile using eq 5, with 2r(pc) as d12. e Estimated from the dipolar splitting of the +2 oxidation state ESR spectrum. (See refs 16b,c.)

distances obtained from the dipolar splitting of the triplet form of the dication diradical (d(ESR), Å ) 30.3/D1/3, where D is the dipolar splitting in Gauss) in making Hab value estimates from the optical spectra.16b,c,17 However, because d(ESR) arises from an experimental measurement which refers to adiabatic energy surfaces, it is an experimental estimate of d12. To the extent that dab and d12 differ, what we did departs from the general practice of using dab estimates for the ET distance since 1967. This paper considers how different these two quantities are predicted to be, and how much they differ from the traditional estimates of charge center separation, employing semiempirical AM1 calculations18 (at the unrestricted HartreeFock (UHF) level) to estimate d12 for several types of nitrogencentered IV compounds. Estimates of Electron-Transfer Distances from AM1 Point Charge Distributions. We shall first consider very simple estimates of d12 for the three +Hy-Ar-Hy examples shown above (see Table 1). For these monocations, we first employ the centroid of atomic point charges (which we will call the center of charge)19 relative to the PH midpoint (r(pc), for point charge) as the basis for estimating d12. Direct calculation of dab does not seem achievable because the bridge obviously interacts with the hydrazine units, and we see no simple way to turn off this interaction in a calculation on an IV compound. An estimate of dab in the spirit of using the metal-metal distance for metalcentered compounds would be to use distance between the midpoints of the NN bonds as dab, using the assumption that the center of charge for the oxidized hydrazine unit lies at the midpoint of the NN bond. However, the aryl bridge has quite different electronic properties from saturated alkyl groups, so it ought to be a better assumption to include the bridge as well as the Hy unit in calculating the center of charge. We therefore consider the center of charge for a monohydrazine model, +Hy-PH-H. It has no Hab because it lacks the second Hy unit, but it can delocalize the positive charge onto the phenyl group as well as the alkyl groups. The center of charge for the AM1-UHF optimized structure lies 0.21 Å from the midpoint of the NN bond and corresponds to r(pc) ) 3.07 Å. We therefore estimate dab for +Hy-PH-Hy using this model compound as 2r(pc) ) 6.14 Å. However, 6.14 Å does not yield a realistic Hab for +Hy-PH-Hy using eq 2. When the Hab it produces is used in eq 4, the calculated kopt is 7.5 to 5.5-fold smaller than that using the 4.6-4.8 Å estimate based on d(ESR). We only know kESR approximately for +Hy-PH-Hy because it is inconveniently large on the ESR time scale for very accurate measurement, but it is a few times larger than that estimated

Figure 1. Plot of r(pc) versus the Ar-NH2 twist of +Hy-PH-NH2, and for comparison, r(pc) values for +Hy-PH-H and +Hy-PH-Hy. The r(pc) values are given in Å.

using 2r(pc) as dab in eq 2. kESR approximately agrees with the kopt calculated using d12(ESR).16b Perhaps a better model compound for +Hy-PH-Hy than +Hy-PH-H could be found, but if a substituent is included at the para position of the bridge of a model system, this system will have an Hab value for electronic interaction of the substituent with +Hy through the bridge. Figure 1 compares the r(pc) values for +Hy-PH-NH having enforced twisting at the C -NH bond 2 Ar 2 with those for +Hy-PH-H and +Hy-PH-Hy. When the NH2, aryl ring twist angle φ ) 90° and through-ring electronic coupling is lost, r(pc) is nearly as large as for +Hy-PH-H, but as this coupling becomes larger when the NH2,Ar overlap is increased, r(pc) drops, following the expected cosφ relationship. Coupling of NH2 with +Hy never becomes as effective as that of Hy because the energy gap is larger for NH2. Figure 1 indicates that estimating the electron-transfer distance from the geometric distance between the charge centers of organic M units is not a good way to proceed. Structural changes five bonds away from the aryl nitrogen cause large changes in the calculated distance, and there appears to be no good way to choose a compound to use as the model that would produce the distance required. The geometric distance between the charge centers of the M groups has almost always been used in calculating Hab using eq 2, but there are apparently problems. One might expect to obtain d12 from r(pc) calculated in the same manner for the IV compound (d12(pc) ) 2r(pc)). The charge center calculated for +Hy-PH-Hy lies nearer the midpoint of the N-CAr bond, and r(pc) ) 2.28 Å. There are four nearly isoenergetic diastereomers calculated for +Hy-PH-Hy. For the sNsB isomer (see Scheme 1 below) we obtain d12(pc) ) 4.56 Å. Use of this estimate together with eq 5 and the optical data yields a dab estimate of 4.90 Å, rather close to the d(ESR) of 4.6-4.8 Å obtained from a compound, and also compatible with the rather imprecisely known kESR for +Hy-PH-Hy. Figure 1 includes the r12(pc) value for +Hy-PH-Hy, which is seen to be considerably smaller than the value obtained for +Hy-PH-H. These calculations indicate that when there is a substantial electronic interaction through a bridge as in this compound, using the center of charge in the +M-B unit as a model to estimate the electron-transfer distance is inadequate.

10026 J. Phys. Chem. A, Vol. 104, No. 44, 2000

Nelsen and Newton

SCHEME 1: Diastereomers of Hy-PH-Hy

Increasing the number of bonds in the bridge to nine in increases r(pc) to 4.21 Å, so a maximum d12 estimate is 8.42 Å, while the distance between the NN bond midpoints is 10.7 Å, and the edge-to-edge NAr-NAr distance is 9.99 Å for the minimum energy diastereomer. Once again, a dab value estimated from the centers of the NN bonds or the edges of the charge bearing unit is larger than dab(pc) estimated from the charge and ring centers and the observed optical spectrum. The difference between dab(pc) and 2r12(pc) is significantly smaller (1.2%) for the smaller Hab BI-bridged compound than it is for the PH-bridged one (7.5%). The observed kESR is compatible with Hab obtained from eq 2 when the experimental dESR estimate from the dipolar splitting of the dication, 8.0 Å, is used, but this is not very different from using dab(pc). The effect of decreasing Hab by increasing twist at the aryl ring hydrazine units upon methylating the p-phenylene bridge, which does not change dNArNAr, is apparently not predicted properly by AM1 calculations. The AM1 value of 2r(pc) for +Hy-DU-Hy is only 0.03 Å larger than for the less twisted, higher Hab +Hy-PH-Hy. Equation 5 predicts a smaller dab(pc) value for +Hy-DU-Hy than for +Hy-PH-Hy because µ12 is significantly smaller. Using the d(ESR) values with eq 2 for both +Hy-PH-Hy and for +Hy-DU-Hy does produce agreement with the experimental kopt values.16b Estimates of d12 for Aryl-Bridged Bis(hydrazines) from Dipole Moment Changes. The interpretation of the chargecenter estimations of d12 using 2r(pc) as discussed above is based on an assumed relationship between a point charge representation of charge densities and the effective ET distance. A more direct approach is to calculate the change in dipole moment, ∆µ12. Cave and Newton used CI to describe the relevant states,4,20 but here we investigate an approximate computational shortcut.21 If the dipole simply switches sign as the charge centroid switches from one end of the molecule to the other upon excitation, which is qualitatively what happens in the symmetrical IV compounds under discussion, eq 7 might be +Hy-BI-Hy

d12(dm) ) 2 µ1(Debye)/4.8032

(7)

expected to be a reasonable approximation for d12 (in Å), where µ1 is the calculated AM1-UHF dipole moment (dm) for the ground state of the IV compound. Using eq 7 is quite similar to using the point charge approach, but accounts for the slightly nonspherical distribution of electrons about the heavy atoms. The dipole moment of an ion depends on the choice of the origin for the calculation, and we have used the center of mass as the origin for the dipole moment calculation. Equation 7 is an approximation because the geometries are slightly different at

the oxidized and reduced M units, but the approximation appears to be quite a good one (see the section on bis(hydrazines) with saturated bridges). It is not clear how to define the electron transfer axis rigorously for localized IV compounds, which have no symmetry because the geometries of the charge bearing units differ somewhat. In all cases, the largest component of the dipole moment vector (µ1) is, as expected, along the long axis of the molecule, which corresponds at least roughly to the electrontransfer axis. Support for adopting the long axis (whose precise definition depends on the detailed structure of the IV system, as indicated in the table footnotes) is provided (for a small sample of the IV systems considered in the present paper) by supplementary CI calculations22 carried out at the INDO level15 for both ∆µ b12 and b µ12. These calculations show that ∆µ b12 and b µ12 are nearly collinear and are also essentially collinear with µ2, the dipole moment the long axis component of b µ1 (as well as b of the final state). They also indicate that the approximation for ∆µ12 entailed in eq 7 is accurate to 10% or better. Finally, these calculations reveal that in some cases b µ1 and b µ2 may have appreciable components perpendicular to the long axis that tend to cancel in ∆µ b12. Some related examples based on the present AM1-UHF calculations are also noted below, and estimates of the effect are provided in the tables by the parameter Rµ, the direction cosine between the long axis component and b µ1. We will use the designation dab(dm) for dab values obtained using using d12(dm) values obtained from eq 5 using d12(dm) values based on eq 7. These dab(dm) values appear in the tables, and we believe they constitute a simple, convenient method for extracting electron-transfer distance estimates from a combination of experimental optical measurements and calculations.23 Diastereomers of similar energy that interconvert by nitrogen inversions and CN bond rotation exist for most of the compounds studied here, and we are especially interested in finding out whether these diastereomers should have significantly different d values. Carbon NMR has shown that at -50 °C, the neutral species Hy-PH-Hy is present as an equal mixture of four slowly interconverting diastereomers,16 shown diagrammatically in Scheme 1. There is a most favorable twist angle between the axes of the aryl ring pπ orbital and the lone pair at the nitrogen attached to the aryl ring (φ ) 33°), but these diastereomers differ in other aspects of the relative orientation of their Hy units. Double nitrogen inversion in the Hy units (the vertical arrows in Scheme 1) is slow on the NMR time scale even at elevated temperatures (single nitrogen inversions would give cis alkyl groups; such conformations are much less stable than ones having trans alkyls for neutral bicyclic hydrazines). Aryl, N rotation (the horizontal arrows) is slow at -50 °C for Hy-PH-Hy. This causes separation of signals for diastereomers that differ by having the twist angle between their NN bonds (χ) nearer 0° (which we will abbreviate sN) or nearer 180° (aN), and the tert-butyl groups may either be on the same side of the bridge (sB) or on opposite sides (aB). The nearly equal amounts of these four diastereomers that are present show that the energy of one Hy unit is not affected by the relative orientation of the other one. The intervalence radical cation oxidation state diastereomers are calculated to also have very similar enthalpies, total range 0.1 kcal/mol (see Table 2). The charge is principally localized in one hydrazine unit, and the geometries at the cationic and neutral hydrazine units are rather different. A more realistic view of the sNsB diastereomer is shown below, with the central bonds for the two types of twist angles (φ and θ) indicated. The X-ray structure of crystalline

ET Estimations from AM1

J. Phys. Chem. A, Vol. 104, No. 44, 2000 10027

TABLE 2: AM1-UHF Twist Angles (deg.) and d Values (Å) for +Hy-PH-Hy Diastereomers sNsB sNaB aNsB aNaB •(sNsB)x-ray f

rel. ∆Hf a

χb

θ+ c

θ0

φ+ d

φ0

dN,N

d12(dm)

Rµ e

0.00 0.05 0.02 0.10

33.0 50.6 129.1 150.9 23.6

157.8 157.6 157.7 157.5 142.4

130.5 130.6 130.4 130.5 147.3

55.8 55.1 55.4 54.7 47.6

32.8 32.6 32.4 32.7 32.6

5.70 5.70 5.70 5.70 5.66

4.47 4.48 4.48 4.47 s

0.996 0.993 0.983 0.980 s

a Unit: kcal/mol. b χ is the twist angle between the two NN bonds. c θ is the twist angle between the nitrogen lone pair axes, calculated assuming the axes bisect the CNC bond angle in a projection down the NN bond. The + and 0 subscripts refer to the radical cation and neutral hydrazine. d φ is the twist angle of the nitrogen lone pair axis, aromatic carbon p orbital axis at the N-Ar bond. e Rµ is the direction cosine between the calculated dipole vector and the charge transfer direction, taken as the vector corresponding to the long axis of the system. The long axis direction was chosen as the line between the aryl nitrogens for the π-bridged systems and that between the midpoints of the NN bonds for the saturatedbridged systems. f From the Supporting Information for ref 16a.

all of the points shown in Figure 2 for the sNsB diastereomer lie within 0.09 Å (0.02%) of their average. Thus despite the presence of four diastereomers and great flexibility at the oxidized hydrazine unit for each of them, all produce nearly the same d12(dm) value. It is important that this be the case for a Hush analysis of the optical data to work well. Similar calculations for other aromatic-bridged bis(hydrazines) in their most stable form are compared with available X-ray data in Table 3. In addition to the compounds discussed above, we include p-xylylene (XY) and 9,9-dimethyl-2,7fluorenylene (FL), the latter being the central bond twist angle

Figure 2. Plot of the change in ∆Hf as a function of forcing an NN+Ar twist angle (φ+) to change. A very small increase in ∆Hf is calculated because NAr, which is nearly planar, inverts without changing the steric interaction of the tert-butyl group with the ring significantly. +Hy-PH-Hy

Ph4B- shows it to be in the sNsB diastereomer, to have φ0 and dN,N quite similar to the calculated values, and

to have a 16° smaller θ+, 17° larger θ0, and 8° smaller φ+ value. However, these compounds are exceptionally flexible at the oxidized Hy unit. Figure 2 shows that there is a double minimum in the ∆∆Hf versus φ+ plot, and that the heat of formation (∆Hf) is calculated to change less than 0.2 kcal/mol for a 60° change in φ+. This occurs because the very flattened NAr nitrogen has only a very small inversion barrier, and the oxidized hydrazine unit changes from having tert-butyl and Ar anti (θ+ large, 158° at the minimum enthalpy point) to syn (θ+ small, 7° at the minimum enthalpy point) near the maximum of the curve (about φ+ ) 125°). This flexibility at the oxidized hydrazine unit (NN+) allows the relative position of the bulky tert-butyl and aryl groups to change little for rather large changes in φ+. The geometry obtained for the neutral hydrazine unit (NN0) is rather insensitive to changes at NN+. As shown in Table 2, d12(dm) is quite similar (4.53(4) Å) for all four optimized diastereomers. Furthermore, it is also insensitive to changes in φ+, because

ω ) 0 analogue of BI. Increasing steric hindrance between the bridge and the hydrazine units by adding methyl groups to the bridge in XY and DU significantly increases the calculated φ values (see Table 3). Diastereomers having the NN bond syn to the methyl-substituted carbon for the XY bridge are significantly less stable than those having the NN bond anti (that is, syn to the CH bond, causing less steric hindrance), and only these diastereomers are calculated to be populated, consistent both with the 13C NMR of neutral Hy-XY-Hy,16b and X-ray data. The most stable diastereomer of +Hy-BI-Hy, sNsB, has a central bond twist angle ω ) 32.7° in its most stable form. The ω ) 0 form (not shown here) is calculated to be 1.1 kcal/mol less stable, and d12(dm) only drops 1% compared to the most stable form. As might be expected because of the similarity in steric interactions, the φ values for the BI- and FL-bridged compounds are calculated to be close to those for the PH bridge. Estimates of d12 for IV Arylamines and TTF Derivatives. Similar calculations for IV compounds having arylamine chargebearing units are summarized in Table 4. Lambert and No¨ll recently published an IV-CT band analysis for the acetylenebridged bis(triarylamine) +An2NPH-C2-PHNAn2 (An ) p-methoxybenzene).24 They used a d of 12.48 Å (the N-N

distance) in extracting H using eq 2.24 We get very close to this value for the N-N distance, but obtain a dab(dm) of 11.4 Å. We also calculated three unsaturated-bridged phenothiazine2-yl (PT) derivatives (R ) Me), as models for the R ) n-Heptyl

10028 J. Phys. Chem. A, Vol. 104, No. 44, 2000

Nelsen and Newton

TABLE 3: Twist Angles (deg.) and ET Distances (Å) for Aromatic-Bridged Bis(hydrazine) Radical Cations (X-ray Data for Comparison Are Marked with • and Shown in Italics) bridge

θ+a

φ+a

φ0a

dN,N

d12(dm)

µ12b

dab(dm)c

d(ESR)d

PH (sNsB) •(1+)[Ph4B-] •+Hy-PH-H[NO3-] XY (aNaB) •(2+)[Ph4B-]2 DU (sNsB) •(1+)[Ph4B-] •Hy-DU-H+[NO3-] BI (sNsB)e •(2+)[SbF6-]2f •+Hy-(PH)2H[SbF6-]g FL (sNsB)

157.8 142.4 152.8 160.1 149.7 164.7 156.7 158.3 157.6 144.5 154.8 157.2

55.8 47.6 59.7 69.6 57.4 71.6 66.2 63.8 54.3 44.0 62.2 57.3

32.8 32.6 s 58.9 s 52.4 50.5h s 36.2 s s 35.0

5.70 5.66 s 5.71 5.64 5.70 5.67h s 9.99 s s 9.72

4.47

4.30

4.82

[4.6-4.8]

4.71

3.08

4.88

5.25

4.58

2.31

4.68

5.66

8.41

3.18

8.51

8.0

8.15

3.79

8.30

[7.5]

a

b

c

See footnotes for Table 2. θ0 is insensitive to both compound and conformation, and is not included. Unit: Debye. Calculated from µ12 in acetonitrile using eq 5 with d12(dm) as d12. d Estimated from dipolar splitting in the ESR spectrum of the triplet bis(cation), and used in our previous work, see refs 16b,c. (Numbers in brackets were estimated from the numbers not in brackets). e Twist angle about the central bond, ω ) 32.7°.f ω ) 35.4°. g ω ) 40.9°. h Average over three conformations of the neutral hydrazine unit present, see ref 16b.

TABLE 4: Comparison of dN,N, d12(dm), and dab(dm), Å, for Intervalence Bis(arylamines) and a Dimeric TTF Derivativea compound +An

d

2N-PHC2PH-NAn2 +PT-C -PTe 2 +PT-(CH) -PTe 2 +PT-PH-PTe +(TTF) f 2

dN,N

d12(dm)

µ12b

dab(dm)c

12.46 12.50 12.19 14.26 8.61g

9.78 9.09 8.75 10.93 6.84

11.40 9.65 6.89 5.60 6.86

10.87 9.94 9.21 11.18 7.41

a PM3 calculations were used for the sulfur-containing PT and (TTF)2 derivatives. b Unit: Debye. Calculated from eq 4 using data from footnotes d to f. c Table 3, footnote b. d See ref 24. e From unpublished work of Daub and Engl. f See ref 25. g The distance between the centers of the S2CdCS2 bonds that was used in calculating Hab (ref 25) is given.

radical cations that have been studied by Ju¨rgen Daub and Raimund Engl (Regensburg, unpublished). We used PM3-UHF calculations, because AM1 appears to perform poorly for sulfurcontaining compounds. These PM3-UHF calculations suffer from massive amounts of spin contamination ( 1.9-2.3), as do AM1 calculations on these systems. Nevertheless, the dab(dm) values obtained are contracted relative to the N-N distances by amounts rather similar to those found for the triarylamines (Table 4). The bis(tetrathiofulvalene) derivative (TTF)2 was included as a representative of the compounds

studied by Lahlil and co-workers.25 The distance between the midpoints of the CdC bonds is 8.6 Å (which we note is a centerto-center instead of an edge-to-edge distance), in contrast to a dab(dm) value of 7.4 Å for this compound. The calculations discussed provide an internally consistent way to estimate the electron-transfer distance for these compounds. Estimates of d12 for Doubly σ-Bridged IV Compounds. The compounds that we consider here with saturated bridges have two links of the same size between the nitrogens, so their NN bonds are nearly exactly perpendicular to the long axis of the molecule. We tabulate the average of the distances between nitrogens linked by the shortest σ chains, dN,N, corresponding to the distance between the midpoints of the NN bonds. Because the π system of the charge-bearing unit contains only these two nitrogens, the edge-to-edge and center-to-center distances are

the same for these compounds. We studied two sorts of double 4-σ-bond links between the dinitrogen units, the bridges abbreviated here as 4T (T is for Tetracyclic) and 4H (H is for Hexacyclic) and one 6-σ-bond linked analogue, 6σ. Bis-

(diazeniums)8 have one additional substituent on each dinitrogen unit, tert-butyl (B) for the compounds considered here, and these substituents may be syn or anti (s or a) relative to each other.

The oxidized trialkyl diazenium unit of the IV radical cation has a NdN bond and the quaternary carbon of the B substituent is coplanar with the C-NdN-C unit. The neutral trialkyldiazenium unit (often called a trialkylhydrazyl) has some twist at the 3e-π NN bond (that is, three electrons in the nitrogen π orbitals perpendicular to the quasiplanar CNNC unit), and the trisubstituted nitrogen is slightly pyramidalized, leading to isomers with the B group directed toward the center of the molecule (which we will call iB0 for inner) or away from the center of the molecule (oB0 for outer), as illustrated for two examples above. The oB0 form is calculated to be only slightly more stable than the iB0 form for the very rigid 4H bridge, but the difference is larger for the more twisted 4T- and 6σ-bridged compounds (see Table 5). The d12(dm) values are slightly larger (1-2%) for the iB0 forms. The syn-4T compound has almost the same average dN,N as the anti one, but a slightly larger d12(dm) value. The 4H compound has a slightly larger dN,N than its 4T analogue because the four-membered ring tilts the dinitrogen units away from each other, but a smaller d12(dm) value. The 4T- and 4H-bridged bis(diazeniums) have kESR values that are too large to be measured,8a but kESR has been determined for both diastereomers of the 6σ-bridged compounds.9 Hydrazines have one additional substituent at each nitrogen. The first bis(hydrazine) radical cation we shall discuss is bis(sesquibicyclic), abbreviated +22/4H/22.8b AM1 yields no twist at either oxidized or reduced hydrazine units of this sort

ET Estimations from AM1

J. Phys. Chem. A, Vol. 104, No. 44, 2000 10029

TABLE 5: Estimated d Values (Å) for Bis(diazenium) Radical Cations with Saturated Bridges compound a+B/4T/B(oB0) (iB0) s+B/4T/B(oB0) (iB0) a+B/4H/B(oB0) (iB0) a+B/6σ/B(oB0) (iB0) s+B/6σ/B(oB0) (iB0)

rel. ∆Hfa ave. dN,Nb d12(dm) 0 0.62 0 0.78 0 0.25 0 0.67 0 0.61

4.82 4.84 4.83 4.84 4.94 4.95 7.16 7.19 7.16 7.19

3.49 3.57 3.65 3.69 3.39 3.43 5.62 5.77 5.62 5.77

Rµc

µ12d dab(dm)

.984 .981 .938 .954 .988 .986 .993 .994 .974 .976

2.63

3.60

3.13

3.66

3.28

3.65

1.07

5.64

1.08

5.64

a Unit: kcal/mol. Relative enthalpies of formation of inner and outer tert-butyl at the slightly pyramidalized nitrogen of the reduced unit, in kcal/mol. b Here, the long axis essentially bisects the two NN bonds. c See Table 2, footnote e. d Unit: Debye. Calculated using the experimental data of ref 8 with eq 4.

(lone pair, lone pair dihedral angle θ ) 0°), but there is still conformational complexity because the alkyl groups may be directed outer or inner in both oxidation states, leading to four stereoisomers, of which the o0o+ form is shown. The relative

energies and d12(dm) values are shown in Table 6. Smaller d12(est) is obtained when the alkyl groups on the oxidized hydrazine are directed i than when they are directed o, and when alkyl groups on the neutral hydrazine are directed o than i. The range in d12(dm) is 9%. Because of their relatively high point group symmetry (Cs for minimum energy equilibrium configuration and C2V for symmetric transition state structure), the oo and ii diastereomers of +22/4H/22 offer a clear understanding of the roles played by the different dipole moment components. For Cs or C2V symmetry, the component of b µ1 parallel to the NN bonds is zero, and we denote the remaining two components as µ| (parallel to the long axis) and µ⊥ (perpendicular to the N4 plane). Table 7 reports the results of three sets of calculations. For the equilibrium structures, the dominant b µ1 component is µ|, which yields the d12(dm) values cited in Table 6 (i.e., 2µ|/4.8032). The value of µ⊥ is, however, seen to be appreciable. Similar b µ1 components and d12(dm) values are found for the minimum energy C2V structure (the d12(dm) values are ∼90% of those for the Cs structures). Due to symmetry-breaking (with respect to C2V) the calculations yield charge densities corresponding to the lower Cs symmetry. Despite the artifactual nature of this symmetry breaking, such symmetry broken wave functions have nevertheless been shown to be useful in modeling ET processes,26 and in fact, the calculated relative ∆Hf values provide rough estimates (actually, diabatic upper limits) of activation energies. Finally, it is possible using the C2V structures to obtain a higher energy SCF solution by constraining the charge density to C2V symmetry.27 These latter charge densities yield µ| ) 0 (as required for symmetrically delocalized states), but with the µ⊥ component little changed relative to the Cs result. This roughly constant µ⊥ (for all entries in Table 7) is not relevant to the ET process of interest and thus should not be included in the implementations of eqs 1, 3, 5, and 7.28,29 When nitrogen substituents external to the bridge are not linked, the neutral hydrazine unit has these substituents directed

on opposite sides of the nearly planar CbrNNCbr unit (θ0 ≈ 130° and insensitive to diastereomer present), but the oxidized hydrazine unit for these saturated compounds is almost always more stable with its substituents on the same sides of the CbrNNCbr unit (θ near 0°), in contrast to the aromatic-bridged bis(Hy) compounds, which are calculated to have large θ for oxidized as well as neutral hydrazine units. Several forms of some tert-butyl, isopropyl, and methyl substituents have been studied experimentally,8 and calculations on the t-Butyl, Methyl (BM) compounds studied appear in Table 6. We abbreviate the conformation of s+BM/4H/BM having the disposition of substituents indicated above as iB+iM+,oB0iM0. We found only large θ alkyl group minima for a+BM/4T/BM. Rather little difference in d12(dm) is predicted between small and large θ forms when the same substituents are present, and we only show dab(dm) values for the most stable diastereomer in Table 6. Only optical analyses are available for 4T- and 4H-bridged BMsubstituted compounds because kESR is too small to measure. Discussion Because of their relative structural simplicity (dominated by the quasirectangular N4 grouping), we commence the discussion by considering the IV compounds with saturated bridges. The separation of the charge-bearing N2 units is essentially the same, whether viewed as edge-to-edge or center-to-center, corresponding to dN,N. By default, we previously used dN,N values in determining Hab values from optical spectra.8,9 The pattern of relative energies calculated for different nitrogen invertomers is complex, depending upon both bridge and alkyl substituents. The sensitivity of d12(dm) for these compounds is largest for the saturated-bridged bis-hydrazines (especially the 4T bridged systems of Table 6, with total ranges 19 and 24% for s and a diastereomers), but the calculated enthalpy differences between invertomers are rather large, and if the calculations were accurate, only the more stable conformations would contribute much to the optical measurements. It should be noted that the dab(dm) values are always smaller than the dN,N values. The range of dab(dm)/dN,N ratios is 0.74 to 0.79 for the bis(diazenium) compounds of Table 5 and 0.72 to 0.85 for the bis(hydrazine) compounds of Table 6. Using dab(dm) values instead of dN,N values would accordingly increase the Hab values obtained from the optical data. The sizes of the changes are not very large, and incorporation of a refractive index correction to the max used in calculating Hab (which was not employed previously)8,9,16,17 would tend to cancel changes caused by decreasing the ET distance used. Because the NN bonds are at approximately 120° angles to the long axis of the aromatic-bridged molecules, the center-tocenter (of the NN bond) distance is significantly larger than the edge-to-edge distances quoted as dN,N in Table 3, with ratios of 1.13-1.15 for the 5-bond-bridged compounds having PH, XY, and DU bridges, and 1.08-1.09 for the 9-bond-bridged compounds having BI and FL bridges. The range of dab(dm)/ dN,N ratios is 0.82-0.85 for the bis(hydrazines) of Table 3 and 0.76 to 0.87 for the bis(arylamines) of Table 4. We used dab estimates that were obtained from ESR measurements on the dication oxidation states (the d12(ESR) values in Table 3) to estimate Hab values for the bis(hydrazines) previously.16,17 The dab(dm) values depend, of course, on the quality of the calculated charge distributions. The AM1 calculations give a reasonable overall account of variation of ET distance with system size but there appear to be problems in accounting for the differential effects of increased twisting caused by introduction of methyl groups in the series PHfXYfDU. As shown in Figure 3,

10030 J. Phys. Chem. A, Vol. 104, No. 44, 2000

Nelsen and Newton

TABLE 6: Estimated Twist Angles (deg.) and d Values (Å) for Bis(hydrazine) Radical Cations Having Saturated Bridges cmpnd. form

rel. ∆Hfa

φ0

ave. dN,Nb

φ+

d12(dm)

Rµc

µ12d

dab(dm)e

+22/4H/22

i+o0 o+,o0 o+,i0 i+,i0

0 0.42 1.81 1.92

0.0 0.0 0.0 0.0

0.0 0.0 0.0 0.0

4.93 4.91 4.99 5.00

3.59 3.98 4.20 3.80

.998 .978 1.000 .966

1.95

3.68 4.06f

iB+iM+,iB0oM0 oB+iM+,iB0oM0 iB+iM+,oB0iM0 iB+oM+,oB0iM0

0 0.23 0.43 0.90

129.4 130.0 130.5 130.2

3.0 8.6 6.5 164.9

a+BM/4T/BM 4.88 4.87 4.87 4.86

3.42 3.96 3.82 4.07

.999 .996 .987

1.66

3.48

.977

s+

iB+iM+,iB0oM0 oB+oM+,iB0oM0 iB+iM+,oB0iM0 oB+oM+,oB0iM0

0 0.65 0.40 0.91

129.6 129.7 130.3 130.2

7.0 0.8 0.8 7.9

BM/4T/BM 4.88 4.87 4.87 4.86

3.48 3.96 3.87 4.33

1.000 .990 .988 .996

1.63

3.54

oB0iM0,oB+iM+ oB0iM0,iB+iM+ iB0oM0,iB+iM+ iB0oM0,oB+oM+

0 0.48 0.43 0.08

130.3 133.7 129.8 129.8

2.2 2.8 0.5 4.4

a+BM/4H/BM 4.98 4.99 5.00 4.99

4.18 3.92 3.78 4.00

.999 .984 .997 .992

1.55

4.23

a Unit: kcal/mol. b The average of the shorter N-N distances across the 4- or 6-σ-bond bridges is used as d c d N,N. See Table 2, footnote e. Unit: Debye. Same as Table 4, footnote b, using the optical data from refs 8b and 8c. e Calculated for the conformation of lowest calculated enthalpy only, using eq 5. f Calculated using d12(dm) for o+,o0, despite the fact that a symmetrical invertomer is unlikely, given the anomalous temperature dependence of kESR (ref 27).

TABLE 7: Dipole Moment Components Calculated for 22/4H/22 Cations diastereomer

sym.

o+,o0 (oo)+

Cs C2V

i+,i0 (ii)+

Cs C2V

charge distributiona

relative ∆Hf (kcal/mol)

µ| (Debye)b

µ⊥ (Debye)c

relative d12(dm)

uns uns sym uns uns sym

0 7.89 20.14 0 7.00 13.48

9.55 9.27 0.00 9.13 8.17 0.00

2.03 2.52 2.65 2.46 2.65 2.63

1 0.97 1 0.89

a

Symmetry of the charge density (unsymmetrical or symmetrical) is defined relative to the point group symmetry. b µ| is the component of µ1 along the long axis. c µ⊥ is the component of µ1 perpendicular to the N4 plane.

Figure 3. Plot of various distances versus the experimental IV chargetransfer transition dipole moment (µ12). The circles represent quantities obtained from the ESR dipolar splitting, and the diamonds ones from AM1 dipole moments. The filled symbols represent d12(dm) values and the open ones dab(dm) values obtained using the µ12 values with eq 5. The paired symbols connected by dots for the +Hy-PH-Hy entries reflect uncertainty in the d12(ESR) estimate.

the experimental d12 estimate, d(ESR), clearly increases significantly in this series (ca. 4.7f5.25f5.66, relative sizes 0.90, 1, and 1.08), but the AM1-calculated d12(dm) values do not

(4.47f4.71f4.58, relative sizes 0.95, 1, and 0.97). Although dab might have been expected to be constant, the dab(ESR) obtained from eq 5 using d12(ESR) and experimental µ12 values clearly is not. The AM1 calculated dab(dm) values are much more nearly constant (4.82f4.88f4.68), but have the DU-bridged compound out of order. It is conceivable that this occurs principally because the DU-bridged compound has more minima (the nitrogens are significantly flatter and twisting is greater), and we might not have yet found suitable conformations. However, the sensitivity of d12(dm) to diastereomer for +Hy-PH-Hy (Table 2) is quite small, with a total range under 0.25%, and the minima found thus far for the DU-bridged compound have quite similar d12(dm) that are anomalously small relative to those for the XY-bridged compound. Rate constants obtained from the optical spectra using eq 2 to calculate Hab show slightly better agreement with the experimental values when the d12 estimate (d(ESR)) is used16 instead of dab(dm), but the difference between using dab and d12 only becomes important as Hab/Eop becomes large, and we do not have a very accurate experimental rate constant for the largest Hab compound, +Hy-PH-Hy, because its k12 value is inconveniently large for our ESR method. It appears that despite having the same charge-bearing units and the structural similarity of the bridges in the series PHfXYfDU, the effective diabatic surfaces are not the same for these compounds, a result that would rationalize the dependence of ET distance upon the introduction of methyl groups on the bridge, as shown in Figure

ET Estimations from AM1 3. Nevertheless, the two state model suffices for predicting kESR from the optical spectrum for these compounds.16c,17 Conclusions The calculations reported here provide an internally consistent way of evaluating dab for organic compounds in which the excess charge on the charge-bearing units is appreciably delocalized, and allow treating systems having π systems perpendicular, parallel, or canted with respect to the electrontransfer direction in a unified manner, which has not been done previously. The use of AM1 calculations with the simple estimate of d12 from the long-axis component of µ1 (eq 7) seems to be appropriate for estimating the electron-transfer distance. The calculations indicate that dab is not very sensitive to the diastereoisomeric mixture expected to be present for most of the compounds studied. It appears to us to be quite reasonable to employ dab(dm) values for extracting Hab from optical data, and to use the smaller refractive index correction f(n),7c which for most of the compounds we studied, will give rather similar values to those obtained by employing d12(dm) values and omitting a refractive index correction. There appears to be a problem with the d12(dm) and hence dab(dm) values obtained for the most twisted compound, +Hy-DU-Hy; the former are smaller than those for +Hy-XY-Hy, and thus not consistent with either experiment or expectation. Acknowledgment. We thank Dr. Timothy Clark and his student Peter Gedeck (Computer-Chemie-Centrum, Erlangen) for providing VAMP programs that carry out dipole moment calculations of ions using the center of mass as the origin, obviously crucial to this work. We thank for partial financial support of this work the Research Corporation under grant RA0269 (S.F.N.) and the Division of Chemical Sciences, U.S. Department of Energy, under grant DE-AC02-98CH10886 (M.D.N.). We thank Prof. Ju¨rgen Daub for sending us unpublished spectra of cations having PT charge-bearing units related to those that appear in Table 4, and Prof. Joseph Hupp (Northwestern) for useful discussions. References and Notes (1) (a) Creutz, C. Prog. Inorg. Chem. 1983, 30, 1. (b) Crutchley, R. J. Prog. Inorg. Chem. 1994, 41, 273. (2) (a) Sutin, N. Prog. Inorg. Chem. 1983, 30, 441. (b) Marcus, R. A.; Sutin, N. Biochim. Biophys. Acta 1985, 811, 265. (3) (a) Hush, N. S. Prog. Inorg. Chem. 1967, 8, 391. (b) Hush, N. S. Coord. Chem. ReV. 1985, 64, 135. (4) (a) Newton, M. D.; Cave, R. J. Molecular Electronics; Jortner, J., Ratner, M., Eds.; Blackwell Science: Oxford, 1997; p 73. (b) Cave, R. J.; Newton, M. D. Chem. Phys. Lett. 1996, 249, 15. (c) Cave, R. J.; Newton, M. D. J. Chem. Phys. 1997, 106, 9213. (d) Newton, M. D. AdV. Chem. Phys. 1999, 106, 303. (5) As it stands, eq 1 refers to the case where b µ12 and ∆µ bab are collinear. In general,4 one requires the projections of these vectors in the charge transfer direction, as discussed below. In the remainder of the paper the absolute value notation is supressed. (6) (a) Jortner, J.; Bixon, M. J. Chem. Phys. 1976, 64, 4860. (b) Ulstrup, J.; Jortner, J. J. Chem. Phys. 1975, 63, 4358. (c) Jortner, J.; Bixon, M. J. Chem. Phys. 1988, 88, 167. (d) Bixon, M.; Jortner, J. J. Phys. Chem. 1993, 97, 13 061. (e) Cortes, J.; Heitele, H.; Jortner, J. J. Phys. Chem. 1994, 98, 2527.

J. Phys. Chem. A, Vol. 104, No. 44, 2000 10031 (7) (a) Gould, I. R.; Noukakis, D.; Gomez-Jahn, L.; Young, R. H.; Goodman, J. L.; Farid, S. Chem. Phys. 1993, 176, 439. (b) Gould, I. R.; Young, R. H.; Mueller, L. J.; Farid, S. J. Am. Chem. Soc. 1994, 116, 8176. (c) Gould, I. R.; Young, R. H.; Mueller, L. J.; Albrecht, A. C.; Farid, S. J. Am. Chem. Soc. 1994, 116, 8188. (d) Gould, I. R.; Farid, S. Acct. Chem. Res. 1996, 29, 522. (8) (a) Nelsen, S. F.; Wolff, J. J.; Chang, H.; Powell, D, R. J. Am. Chem. Soc. 1991, 113, 7882. (b) Nelsen, S. F.; Chang, H.; Wolff, J. J.; Adamus, J. J. Am. Chem. Soc. 1993, 115, 12 276. (c) Nelsen, S. F.; Ramm, M. T.; Wolff, J. J.; Powell, D. R. J. Am. Chem. Soc. 1997, 119, 6863. (9) Nelsen, S. F.; Trieber, D. W., II.; Wolff, J. J.; Powell, D. R.; RogersCrowley, S. J. Am. Chem. Soc. 1997, 119, 6973. (10) Karki, L.; Hupp, J. J. Am. Chem. Soc. 1997, 119, 4070. (11) (a) Paddon-Row, M. N.; Oliver, A. M.; Warman, J. M.; Smit, K. J.; de Haas, M. P.; Oevering, H.; Verhoeven, J. W. J. Phys. Chem. 1988, 92, 6958. (b) Warman, J. M.; Smit, K. J.; Jonker, S. A.; Paddon-Row, M. N.; Oliver, A. M.; Kroon, J.; Oevering, H.; Verhoeven, J. W. J. Phys. Chem. 1991, 95, 1979. (12) (a) Oh, D. H.; Boxer, S. G. J. Am. Chem. Soc. 1990, 112, 8161. (b) Oh, D. H.; Sano, N.; Boxer, S. G. J. Am. Chem. Soc. 1991, 113, 6880. (c) Bublitz, G. U.; Boxer, S. G. Annu. ReV. Phys. Chem. 1997, 48, 213. (13) Karki, L.; Lu, H. P.; Hupp, J. T. J. Phys. Chem. 1996, 100, 15 637. (14) (a) Shin, Y. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Am. Chem. Soc. 1995, 117, 8668. (b) Shin, Y. K.; Brunschwig, B. S.; Creutz, C.; Sutin, N. J. Phys. Chem. 1996, 100, 8157. (15) (a) Zerner, M. C.; Loew, G. H.; Kirchner, R. F.; Meuller-Westerhoff, U. T. J. Am. Chem. Soc. 1980, 102, 589. (b) A comprehensive semiemperical SCF/CI package written by M. C. Zerner and co-workers, University of Florida, Gainesville, FL. (16) (a) Nelsen, S. F.; Ismagilov, R. F.; Powell, D. R. J. Am. Chem. Soc. 1996, 118, 6313. (b) Nelsen, S. F.; Ismagilov, R. F.; Powell, D. R. J. Am. Chem. Soc. 1997, 119, 10 213. (c) Nelsen, S. F.; Ismagilov, R. F.; Gentile, K. F.; Powell, D. R. J. Am. Chem. Soc. 1999, 121, 7108. (17) Nelsen, S. F.; Ismagilov, R. F.; Trieber, D. W., II Science 1997, 248, 777. (18) (a) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F.; Stewart, J. J. P. J. Am. Chem. Soc. 1985, 107, 3902. (b) AM1 calculations in Madison were carried out using various versions of Clark’s program VAMP. Rauhut, G.; Chandrasekhar, J.; Alex, A.; Steinke, T.; Clark, T. VAMP 5.0, Oxford Molecular, Oxford, U.K., 1994. (19) This centroid is a vector quantity, which is essentially parallel to the vector defined by the two para carbons of the PH bridge. The centroid for each coordinate direction is the sum of the net charge times the coordinate value for each atom. (20) Cave, R. J.; Newton, M. D.; Kumar, K.; Zimmt, M. B. J. Phys. Chem. 1995, 99, 17 501. (21) We thank Joseph Hupp (Northwestern University) for discussing this possibility with us. (22) M. D. Newton, to be published separately. (23) The combined use of calculated vacuum results for d12 and experimental solution data for the other optical parameters provides a useful approximate procedure for estimating dab to the extent that solvation effects on the various quantities are relatively small. (24) Lambert, C.; No¨ll, G. Angew. Chem., Int. Ed. Engl. 1998, 37, 2107. (25) Lahlil, K.; Moradpour, A.; Bowlas, C.; Nenou, F.; Cassoux, P.; Bonvoisin, J.; Launay, J. P.; Dive, G.; Dahareng, D. J. Am. Chem. Soc. 1995, 117, 9995. (26) Newton, M. D. Chem. ReV. 1991, 91, 767. (27) Nelsen, S. F. J. Am. Chem. Soc. 1996, 118, 2047. (28) A comment is in order concerning the definition of the charge transfer direction, generally taken as that given by ∆µ b12, as discussed in the text in connection with eq 3. When the latter is zero (as in the C2V density cases displayed in Table 7, see ref 29), a more general definition is required. We propose using in this special limiting case the direction of the deriVatiVe of ∆µ b12 with respect to the reaction coordinate (generally taken as the vertical diabatic energy gap (see ref 30), a quantity which may be straightforwardly obtained from a computational model). (29) Table 7 displays only b µ1 values, but very similar results in the symmetric charge density C2V case are expected for b µ2 (i.e. µ| ) 0 and µ⊥ essentially the same as for b µ1), yielding the result ∆µ b = 0, an expectation borne out by the supplementary INDO/CI results noted above. (See ref 22.) (30) Zusman, L. D. Chem. Phys. 1980, 49, 295.